e-learning technologies on the semantic web rachid benlamri chair, dept. of software engineering...
TRANSCRIPT
e-Learning Technologies on the
Semantic Web
Rachid BenlamriChair, Dept. of Software Engineering
Lakehead UniversityCanada
For Contact Email: [email protected] King Abdulaziz University - Joint Supervision Program – Oct. 24-29, 2009 Jeddah – KSA
3. How it all fits together?Case Study: Context-Aware
M-Learning
3. How it all fits together?Case Study: Context-Aware
M-Learning
4. Conclusions Demos
4. Conclusions Demos
1. MotivationHistorical view,
Problems, Goals & Vision
1. MotivationHistorical view,
Problems, Goals & Vision
Outline
2. What does Semantic Web bring to e-learning?
Semantic Markup: LOM, RDF, Ontology Rule-Markup: Rule-ML, SWRL,
Web Services, Web Agents
2. What does Semantic Web bring to e-learning?
Semantic Markup: LOM, RDF, Ontology Rule-Markup: Rule-ML, SWRL,
Web Services, Web Agents
Objectives
This presentation is intended to provide you with– research motivations in Semantic Web-based
technologies– an understanding of the basic requirements of
service-oriented learning systems– an understanding of what Semantic Web can bring to
e-learning – methods for context modeling and management for
e-learning – a system prototype for context-aware mobile-learning– a system demo for collaborative/social learning
(Unite Software)
3
Slid
e 4
4
Part 1
Motivation
Historical view Problems
Goals & Vision
Part 1
Motivation
Historical view Problems
Goals & Vision
Traditional Learning
Multimedia-Based Learning
e-Learning History
Early Web-Based Learning
Multimedia-Based Learning
e-Learning History (Cont’d)
WWW
Markup
Early web: Simple web pages, Word, Excel, pdf files managed by simple databases of learning
content (markup)
Early Web-Based Learning
Web-based e-Learning Systems
e-Learning History (Cont’d)
WWW
Learning Management Systems: WebCT, Blackboard, Domino, etc. (markup and XML)
Mark-up + XML
Time- and location-dependent Learning aimed at mass participation
Educator-driven Linear access
Learning Object based Systems
Web-based e-Learning Systems
e-Learning History (Cont’d)
Learning Object Systems: SCORM, ADL, IMI, OCPI, etc., (XML-based Metadata + Web Services + Standards )
XML – LOM -SWDL- SOAP
Sequence Link
Correlation Link
LO
LO
LO
LO
LO
LO
LO
LO
LO LO LO
LO
LO
LO LO
LO
LO
Mandatory Learning Object LO
Secondary Learning Object
LO
LO LO LO LO LO LO LO
Web Services
Discover Share
Reuse Interoperability
Can access any LOBuild learning paths out of LOs
LO-based metadata for indexing and retrieval
Learning Objects based Systems
Semantic Web-based e/m/u-Learning Systems
Vision: e-Learning on the Semantic-Web
Confluence of enabling technologies: Web Agents, Ubiquitous Computing, Ontologies, Web Services, and Open Standards
WSDL-SOAP
Sequence Link
Correlation Link
LO
LO
LO
LO
LO
LO
LO
LO
LO LO LO
LO
LO
LO LO
LO
LO
Mandatory Learning Object LO
Secondary Learning Object
LO
LO LO LO LO LO LO LO
Web Services
Discover Share
Reuse
Agents
OWL-SWRL
Semantic Web
Reasoning
Adapt to Context
……
…
Ontologies
Internet 2
Interoperability
Scalable Service Oriented Systems
Confluence of Enabling TechnologiesOntologies
E-learining onSemantic Web
Agents
Web Services
Knowledge Management
Reasoning Tools
Interoperability Service discovery Service invocation
Shared conceptualizations Explicite Knowledge spec. Semantic Annotation
Contextualization Community creation Proactivity Service composition
Semantically well-defined context-aware dynamic services
Information-Transfer
based Learning
• Information transfer– Online repositories
– Teachers produce and learners consume
• Passive learners – Learner cannot impact the learning process
• No Personalization– Uniform Learning
– All learners learn same way
• No contextualization – System is unaware of learner situation
Knowledge-Transfer
based Learning
• Knowledge transfer– Knowledge-driven framework
– Semantically annotated content– Sharing, reusing, and reasoning
• Active learners – Learner has impact on the learning
process
• Personalized Learning– Tailored to the learner’s needs,
background, and learning style
• Contextualized Learning– Adapted to learner’s situation (current
activity, task at hand, and surronding environment)
Learning Models: How we Conceive the Learning
• Contextualised learning– Understanding of concepts through direct experience of their
manifestation in realistic contexts (e.g. providing access to real world data)
• Social learning– User’s mental processes are influenced by social and cultural
contexts• Collaborative learning
– More than a simple information exchange – peers interactions, conversation tracks, knowledge reconstruction
• Personalised learning– Guarantee the learner to reach a cognitive excellence through
different learning path tailored to learner’s needs and preferences
[Ritrova 2005]
13
Requirements of future e-Learning Systems
Emerging web technologies provide new exciting horizons for building smart distributed e-learning environments
• Semantic markup and reasoning– Learning resources from different sources can be linked to
commonly agreed ontologies– Powerful semantic querying to build personalized learning paths– Open standards for resource sharing and reuse
• Service orientation– Learning is conceived as support services – The learning process is enabled by composing services
• Context-awareness – Ability to recognize learner’s current context (activity, place, device)
14
Requirements of future e-Learning Systems(2)
• Learner-centric– Dynamically optimized learning, on demand, and on a learner's
schedule and context. – Appropriate semantic annotation will still define pedagogical
sequences in the learning hierarchy (core, prerequisites, electives, etc...)
• Community and collaboration based– Is central for all kinds of formal and informal learning – Autonomous and dynamic creation of communities
• Dynamicity– the learner can influence the process– the social and context aspects influence the learner
15
Requirements of future e-Learning Systems(3)
• High demand of interoperability– access to resources on heterogeneous environments
• Security and trust– Confidentiality, privacy, copyright issue, etc...
• Simplify– Learning content authoring– Application Development– Metadata exchange
What Semantic Web Brings to e-Learning
Part 2
What does Semantic Web bring to e-Learning?
Learning Object Metadata (LOM)
Semantic Markup (RDF, RDF-S, OWL, OWL-S)
Rule Markup Languages (Rule-ML and SWRL)
Web Services & Web Agents
Part 2
What does Semantic Web bring to e-Learning?
Learning Object Metadata (LOM)
Semantic Markup (RDF, RDF-S, OWL, OWL-S)
Rule Markup Languages (Rule-ML and SWRL)
Web Services & Web Agents
Semantic Web - Definition
The Semantic Web is an extension of the current web in which information is given well-defined meaning, better enabling computers and people to work in co-operation.
[Berners-Lee et al.,
2001]
Semantic Web Layers(T. Berners-Lee et al.)
First Step Towards SW Learning:Augmenting Learning Objects with Metadata
Learning Object : A re-usable chunk of information used as a modular building block for e-Learning content.
Metadata (Data about data)Structural data about Learning Objects (LOs), to augment them with a meaningful classification so that they can be easily reused, transformed, accessed, etc., in order to gain more information on demand.
IEEE LTSC LOM (Learning Object Metadata)
XML-based Metadata
Role of Metadata• SW-techniques allow you to add metadata to distributed resources just like
html allows you to link to such resources.
• Metadata allows Learning Management Systems (LMS) to:– Annotate– Find– Select– Retrieve– combine– use/re-use, and – share
learning content on the Web
• Metadata is not bound to a fixed schema. You may invent a description format of your own and add personal annotation
Sample of Metadata
• The display type of a device• The topic of a of a lecture• The size of a learning resource• The author of a learning resource• The operating system to execute a
program
Metadata Standards
• Often, standards go through three stages:– Technical Specifications
such as AICC, IMS, ARIADNE– Implementation Reference Models
such as ADL SCORM (sharable content object reference model)
– Accredited Standardssuch as IEEE LOM (Learning Object Metadata)
LO-based Learning SystemsThe SCORM Model
Source from ADL’s SCORM
Is Metadata Enough for Sharing Resources on the SW?
• No metadata standard covers every aspect of life
• They just provide information about the packaged learning content
• Different communities develop different vocabularies (ontologies), but they should be able to use each others
• XML is not good at describing vocabulary • There is a need for a language with semantic
markup to describe shared vocabularies
XML• XML documents are labeled trees • Storage is done just like an n-ary tree (DOM)• Tree element = label + Attribute/Value + content• Document Type definition (DTD): Simple grammar (regular
expressions) to describe legal trees (XML-Schema )• It says what elements and attributes are required or optional.
course
Exams ProjectsLectures
MidTerm Final
<course Name=“...”><Lectures>...</Lectures><Exams> <MidTerm>...</MidTerm> <Final>...</Final>
</Exams> <Projects>...</Projects> </course>
Why not use XML to represent ontologies?
XML: Limitations for Semantic Markup
• XML is a universal metalanguage for defining markup• It provides a uniform framework for interchange of data and
metadata between applications• However, XML is unspecific:
- No predetermined vocabulary
- No semantics for relationships
• XML does not provide explicit semantics (meaning) of data– E.g., there is no intended meaning associated with the nesting of
tags– It is up to each application to interpret the nesting.
• XML is not for sharable Web-resources on a broad scale
• So, there is a need for other form of semantic markup
Resource Description Framework (RDF)
for Semantic Markup • RDF provides metadata about Web resources• Basic building block:
Subject -> Predicate -> Object triples– subject is the focus of the statement– predicate describes a property of the subject– property value is the object.
• So, RDF keeps meta-data external to objects• It has an XML syntax• Chained triples form a graph (semantic net)
http://flash.lakeheadu.ca/~rbenlamr
site-owner
Benlamri http://flash.lakeheadu.ca/pres.pdfauthor
28
RDF’s Resources
• Every resource has a URI, a Universal Resource Identifier
• A URI can be – a URL or – unique identifier
• We can think of a resource as an object, a “thing”. So, RDF URI’s can refer to anything and not just digital resources (e.g. lecturer, author, student, device, etc.)
• So, RDF, is extendable and doesn’t require rigid meta-data structures or proprietary standards or fixed vocabularies
29
Properties• Properties describe relations between
resources– e.g. “author”, “prerequisite”, etc.
• Properties are also identified by URIs (kind of resources)
• Advantages of using URIs:– Α worldwide, unique naming scheme– Reduces the problem of distributed data
representation
30
Statements
• Statements assert the properties of resources• A statement is an subject-predicate-object
– It consists of a resource, a property, and a value
• Values can be resources or literals – Literals are atomic values (strings)
• The triple (x,P,y) can be considered as a logical formula P(x,y)– Binary predicate P relates object X to object Y – RDF offers only binary predicates (properties)
What does RDF Schema add?
• Defines vocabulary for RDF• Organizes this vocabulary in a
typed hierarchy• Class, subClassOf, type• Property, subPropertyOf• domain, range
RudiYork
Person
PhDStud Professor
subClassOfsubClassOf
type
hasSuperVisordomain range
type
hasSuperVisor
[Steffen Staab 2006]
32
Basic Ideas of RDF Schema
• RDF is a universal language that lets users describe resources in their own vocabularies – RDF does not assume, nor does it define
semantics of any particular application domain
• The user can do so in RDF Schema using:– Classes and Properties– Class Hierarchies and Inheritance– Property Hierarchies
RDF(S) Contributions to Semantic Markup
• RDF provides a foundation for representing and processing metadata
• RDF has an XML-based syntax to support syntactic interoperability.
• Next step up from plain XML:– (small) ontological commitment to modeling
primitives– possible to define vocabulary
• So, RDF & RDFS allow incremental building of knowledge, and its sharing and reuse
• However, many desirable modelling primitives are missing – no precisely described meaning– no inference model
34
Limitations of the Expressive Power of RDF-S
• Example: Boolean combinations of classes– Sometimes we wish to build new classes by
combining other classes using union, intersection, and complement ( e.g. person is the disjoint union of the classes male and female)
• Therefore we need an ontology layer on top of RDF and RDF Schema to offer:– explicit, formal conceptualizations of domain models– efficient reasoning support – a formal semantics – sufficient expressive power
[G. Antoniou & F.Harmelen]
Ontology
Ontologies enable a better communication between Humans/Machines
Ontologies standardize and formalize the meaning of words through concepts
Ontology in PhilosophyOntology is a branch of philosophy that deals with the nature and the organization of reality
Ontology deals with questions such as:
What characterizes being?Eventually, what is being?
“ People can‘t share knowledge if they do not speak a common language.“
[Davenport & Prusak, 1998]
Ontology is a formal Specification of a shared conceptualization of a domain of interest [Gruber 93]
Formal Specification of
Conceptualization
…
…
…
Concepts
Domain of
Interest
of
Reasoning+
Processable
Group of
sharedPeople
Web agentsApplications
Ontology
Services
cooperation
37
Ontologies for e-Learning
• In e-learning we distinguish between two types of knowledge (ontologies): – Operational Knowledge
• Learner ontology• Activity ontology• Environment ontology( device + environment)
– Domain Knowledge • Content • Pedagogy• Structure (navigation)
Example of Learner Ontology
Standards: RDF(S); OWL (Web Ontology Language)
R
D
D
R
D
R
Object Property
ConductedLearningActivity
Object Property
HasSurroundingEnvironment
Data Property
HasUserName
HasCovered
Class
Learner
Class
Learning Activity
External
XSD: String Class
Language
Class
Learning Resource
Class
Environment
Class
Concept Object Property
ConsumedLearningResource
D
R
D
R
D
Object Property
Data Property
HasPassword R
R
D
HasLearningGoal
Object Property
D
R
Data Property
HasLearningTime
External
XSD: Date_Time
R
D
PreferredLanguage
Object Property
Class
Location
LocatedIn
Object Property
R
Loops in C++
Learning Resource InstanceOf
Property
InstanceOf
Property
English
Learning
Resource
39
Reasoning about Knowledge in Ontology
• Class membership – If x is an instance of a class A, and A is a subclass of
B, then we can infer that x is an instance of B
• Equivalence of classes – If class A is equivalent to class B, and class B is
equivalent to class C, then A is equivalent to C, too• Consistency
– X instance of classes A and B, but A and B are disjoint– This is an indication of an error in the ontology
40
Reasoning about Knowledge in Ontology (2)
• Reasoning support is important for– checking the consistency of the ontology and the knowledge– checking for unintended relationships between classes– automatically classifying instances in classes
• Checks like the preceding ones are valuable for – designing large ontologies, where multiple authors are involved– integrating and sharing ontologies from various sources– Dealing with operational knowledge and domain knowledge in
an efficient way (i.e. context-aware personalized learning)
[G. Antoniou & F.Harmelen]
41
Web Ontology Language (OWL)
• OWL is a knowledge representation language to model ontologies and reason about their embedded knowledge
• OWL is based on formal semantics• Semantics is a prerequisite for reasoning support• Semantics and reasoning support are usually provided
by – mapping an ontology language to a known logical formalism– using automated reasoners that already exist for those
formalisms• OWL is (partially) mapped on a description logic, and
makes use of reasoners • Description logics are a subset of predicate logic for
which efficient reasoning support is possible[G. Antoniou & F.Harmelen]
42
OWL Standard
• Three OWL versions with different expressive power and designed for different requirements:– OWL FULL (fully expressive, not fully decidable)– OWL DL (less expressive, Description logic )– OWL Lite (restricted power of expressiveness)
• OWL is a knowledge representation language with rich modeling primitives: – Classes with data & object properties– Inverse and equivalence properties– Property and cardinality restrictions– Boolean combinations– Enumerations, etc…
43
SW and Rules Markup LanguagesRule-ML & SWRL
• The Semantic Web approach is to express the knowledge in a machine-accessible way using one of the Web languages (RDF/S, OWL, etc.)
• Then, extend these knowledge representation languages with rule markup languages, also expressed in XML-like syntax
• Basic Idea: shared reasoning is enabled through exchange of rules across different applications
44
Web Services – Contribution of Semantic Web Technology
• Web Service: service based, aiming to provide interoperability among distributed loosely coupled components
• Use machine-interpretable descriptions of services to automate:• discovery, invocation, composition and monitoring of
Web services• Share web services across applications (e.g. use
of Web Service Description Language - WSDL)• Web agents can compose simple web services
into complex web services
45
Web Services and OWL-S
• Applications should be able to employ a set of basic classes and properties by declaring and describing services: ontology of services
• OWL-Schema can be used for developing an ontology language for Web services
• It can be viewed as a layer on top of OWL
[G. Antoniou & F.Harmelen]
46
Three Basic Kinds of Knowledge Associated with a Service
• Service profile– Description of the offerings and requirements
of a service– Important for service discovery
• Service model– Description of how a service works
• Service grounding– communication protocol and port numbers to
be used in contacting the service
Web Agents – Contribution of Semantic Web Technology
• Software agents may represent dynamic entities in a Learning Management System
• Shared ontology(es) are used for mutual understanding of terms and embedded knowledge
• LMS Facts w.r.t. learners, learning process, environment, etc., and subsequent LMS decisions are represented by statements
• Information is exchanged between Agents in a Markup language
• Agent negotiation strategies are described in a logical language
• Agents decide about next course of action through inference, based on negotiation strategy and current facts
Web Agents – Contribution of Semantic Web Technology (2)
Slid
e 4
9
Web
49
Part 3
How it all fits together?
Case Study
Context-Aware M-LearningOn the Semantic Web
Part 3
How it all fits together?
Case Study
Context-Aware M-LearningOn the Semantic Web
Yesterday: Gadget Rules
Cool toys…
Too bad they can’t talk to each other…
[Harry Chen]
Today: Communication Rules
Sync. Download
. Done.
Configuration? Too much
work…
[Harry Chen]
Tomorrow: Services Will Rule
Thank God! Pervasive
Computing is here.
[Harry Chen]
Context-Aware Systems
• “Systems that can anticipate the needs of users and act in advance by “understanding” their context.” [Chen 2003]
• “Systems that are able to adapt their operations to the current context without explicit user intervention. Context –awareness here aims at increasing usability and effectiveness by taking environmental context into account.”
[Dustdar 2006]
Context - Definition
“Context is any information that can be used to characterize the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and applications themselves.”
[Dey and Abowd’s ]
Ontology–based Context Models• Need for knowledge sharing
– Knowledge sharing => Global Ontology spaces to enhance cross-domain understanding
• Use OWL as a meta-language to define other languages that are used in context-aware systems– Policy languages for privacy and security– Content languages for agent communications
• Use the ontology semantics of OWL+ SWRL (SW Rule Language) to reason about context– Deduce context knowledge that can’t be directly
acquired from the sensors– Detect and resolve inconsistent knowledge that
results from imperfect sensing – Infer new knowledge for service adaptation
Context Awareness Pyramid
Context Acquisition (World)
Context Perception
Context Understandin
g & Usage
Expre
ssiven
es
s
Sensory Data
Context Information
Semantics/Understanding/Insight
Context Awareness Pyramid Conceptual Model
58
Raw-Context DataContext Sources
Atomic Context Sensor
Atomic Context Sensor
Atomic Context Sensor
Atomic Context Sensor
Context Sensing Layer
Context Reasoning and Service Adaptation
Context Inference Context LearningContext Perception Layer
Context Understanding and Usage Layer
Context Computation
Raw-Context Data
Raw-Context Data
Raw-Context Data
What is Atomic Context ?
• External (Sensed)– Context that can be measured by hardware sensors
• location, light, sound, temperature, air pressure, blood pressure, heart rate, etc.
• Internal (Retrieved from UI / Profiled )– Mostly specified by the user or captured monitoring
the user’s interaction• the user’s goal, tasks, work context, business
processes, etc.– Retrieved from profiles of users, tasks, schedule,
etc.
Sensing Atomic Context• Sensors
– Physical sensors• sensor, camera, microphone, accelerometer, GPS,
thermometer, biosensors– Virtual sensors
• From software: browsing an electronic calendar, a travel booking system, emails, mouse movements, keyboard input
– Logical sensors• Combination of physical and virtual sensors with additional
information from databases: analyzing logins, mapping fixed devices to location information, etc.
• Sensory Data Capture– Drivers and APIs– Query functionality (ex: getPosition())– Etc.
[Dustdar 2006]
Modeling Atomic Context:Context Atom Attributes
– Context type (Nature of context)
– Context value (Quantized / non quantized( boolean, literal) )
– Description (Symbolic description for high level reasoning)
– Time stamp (at acquisition time)
– Source (Sensor ID)
– Confidence (Truth probability)
Context Awareness Pyramid
Context Acquisition (World)
Context Perception
Context Understandin
g & Usage
Expre
ssiven
ess
Sensory Data
Context Information
Semantics/Understanding/Insight
Data Pyramid
Context Perception– Involves monitoring and simple recognition– Links sensors to network and network-centric
services. – Enabled through:
• Reasoning and interpreting• Computation
– Extraction and quantization operations– Aggregation or compositing of atomic context
elements• Learning techniques
– Statistical methods where training phase is required
– To deal with ambiguous context
Context Awareness Pyramid
Context Acquisition (World)
Context Perception
Context Understandin
g & Usage
Expre
ssiven
ess
Sensory Data
Context Information
Service layer
Semantics/Understanding/Insight
Data Pyramid
Context Understanding and Usage
• Semantic Analysis of Context Involves: – Reasoning: support of axiomatic rules (e.g.
expressed in SWRL)– Interpretation– Recognition, and – Evaluation
• Goal: – Awareness of the likely evolution of a
situation, – Identify its possible/probable future states and
events– Service adaptation to deal with the new
situation
Context Awareness in M-learning
66
Challenges & Contributions• Challenges : Today’s Learning Technology is characterised by:
Passive components rather than active components Separate use of contextual information on users, environment, and
services Lack of techniques for modeling and reasoning with the global
context
• Contributions Proactivity: context sensing, prediction, & management Knowledge Representation: Ontological approach for context
integration and aggregation Knowledge Management: Hybrid approach that may combine
probabilistic, logic, and rule based reasoning
67
68
System Architecture
Context
Sensing
&
Acquisition
Ontology Reasoning
System
Centric
Adaptations
Learner
Centric
Adaptations
Ontology Management
Global Ontology Space
LO Repository 1
Learner Profile Repository
Activity Repository
Device Repository
Environment Repository
LO Repository N
……
LO Repository 2
Run Time Environment Surrounding
Environment
Nomadic
Learners
Service
Discovery
A ge n t
69
Activity Context Learner Context Extends
Leaner-Centric Context
Device Context Extends
System-Centric Context
Context Integration & Aggregation
Personalization
Service Adaptation
Execution Profile
Environment Context
Context Aggregation
Global Ontology Space
70
Learner Ontology
Domain Ontology
Device Ontology
Activity Ontology
Environment Ontology
Concept
LearnerHasCovered
Query
HasKeyword
Environment
HasSurroundingEnvironment
Learning Resource
ConsumedLearningResource
Learning Activity
ConductedLearningActivity
Device
UsedDeviceIsMappedTo
MakeQuery
HasLearningGoal
LocatedAt
Location
HasLocatiom
71
Domain Ontology
D
R
R
D
R
R D D D
D R
D
R
Object Property
HasKeyword
d
Object Property
Isa
Object Property
HasCovered
d
Object Property
IsMappedTo
Object Property
HasPrerequisite
e
Object Property
HasPart
Object Property
HasNecessaryPart
Class
Concept
Class
Query
Class
Learning Resource Class
Learner
R D
Object Property
ExpressedIn
Class
Language
Object Property
HasType Class
Media Type D
D
R
R
Object Property
HasLearningGoal
D
R
External
XSD: Time Data Property
LearningTime
R
72
R
D
D
R
D
R
Object Property
ConductedLearningActivity
Object Property
HasSurroundingEnvironment
Data Property
HasUserName
HasCovered
Class
Learner
Class
Learning Activity
External
XSD: String Class
Language
Class
Learning Resource
Class
Environment
Class
Concept Object Property
ConsumedLearningResource
D
R
D
R
D
Object Property
Data Property
HasPassword R
R
D
HasLearningGoal
Object Property
D
R
Data Property
HasLearningTime
External
XSD: Date_Time
R
D
PreferredLanguage
Object Property
Class
Location
LocatedIn
Object Property
R
Learner Ontology
73
Environment Ontology
R
Data Property
D
R
D
D
D D
R
R
Object Property
HasLocation
Data Property
HasBandwith
HasWirelessNetwork
External
XSD: Float
External
XSD: Date_Time
Class
WirelessNetwork
D
R
R
Object Property
Class
Learner
Class
Environment
Class
Location
Object Property
HasSurrroundingEnvironment
D
SensedAt
External
XSD: Boolean Class
WirelessNetworkT ype
Data Property
IsSecured Object Property
HasWirelessNetworkType
R
D
Object Property
LocatedAt D R
74
R
*1
*1
D
R
R
R
D D
D
R R
R
D
D D Object Property
HasHardwareProfile
Object Property
HasDisplayType
Data Property
HasMaxBandwidth
Object Property
HasSofwareProfile Class
Device
Class
Learning Activity
Class
Hardware Profile
External
XSD: Float
Class
Display Type
Class
Software Profile
Object Property
Object Property
HasNetworkAdaptor
UsedDevice
External
XSD: Integer
Object Property
HasDeviceType
Class
Device Type
Object Property
HasKeyboardType
e
Class
Keyboard Type
Data Property
AvailableMemory
External
XSD: Integer
Class
Network Adaptor
Data Property
HasScreenLength
Class
Network Protocol
Object Property
HasNetworkProtocol
Class
S/W Application
Object Property
RunApplication
Class
Language
Object Property
SupportLanguage
Class
Media Type
Class
OS Type
Object Property
HasOS
R
D
R R
R
D D
D
R D R
Object Property
SupportMediaType
D
R
D
D
R
*1
*1 *1
D
Data Property
HasScreenWidth
R
Device Ontology
75
D D
R
D
R
D
R
Object Property
Data Property
ConductedLearningActivity
Object Property
UsedDevice
Begin-time Data Property
HasActivityId
Object Property
MakeQuery
Class
Learning Activity
External
XSD: Date_Time
External
XSD: Integer Class
Learner
Class
Concept
Class
Device
Class
Query
Object Property
HasKeyword
Data Property
End-time
D
R
D
R
D
R
R
Activity Ontology
76
Subject Domain Ontology:C++ Programming
77
1: Login
successful?
User
2: Formulate
query
3: Retrieve subject –
domain ontology
4: Infer query related
concepts
5: Infer metadata using
perceived
system-centric context
6: Discover & filter out
learning resources using
metadata generated in step
(4) and (5)
7: Infer learner’s expertise in
subject-domain and use it to
generate learning sequence
8: Generate
personalized learning
path
9: Update
learning expertise
10 Terminates
learning session
Yes No
<< New interaction>>
<<Logout>>
Ontology Reasoning
<<New Query>>
User
Web BorrowerHTML
WAP BorrowerWML
User
Sensing Type of Network Connection & Maximum Connection
Speed
Sense & Update Actual Network
Bandwidth
Compute Maximum
Resource Size
Adapt Retrieved Resources to
Device Features
Infer Current Bandwidth
Infer Media Type
Global Ontology Space
Device Repository
Environment Repository
System Centric Adaptation
<<Login>>
78
System-Centric Adaptations
Inferring Current Bandwidth• Infer Current Bandwidth based on
Previously sensed bandwidth Fuzzy Membership Function SWRL Rules to Compute Truth Values
79
L6 L5
L4 L2
L3
L1
Low
1
0.2
0.8
B
Lowband Mediumband Highband B* Maxband
μA(B)
0
Medium High
Fuzzifier Fuzzy Inference Engine Defuzzifer
Fuzzy Rule Base
B Z
Resource-Size & Media-Type Selection
• Resource Size & Media Type Rules
• Resource Size Computation
80
)()()(
)(*arg)(*)(**
*arg
**
*arg
**
3
1
*
*3
1*
BBB
BeSizeLBMediumSizeBSmallSize
B
BZZ
eLMediumLow
eLMediumLow
iA
Ai
i
i
i
Knowledge base for the fuzzy logic system:
• if bandwidth is Low then set file size to Small.• if bandwidth is Medium then set file size to Medium• if bandwidth is High then no file size restriction
Small
1
Z
SmallSize MediumSize MaxSize LargeSize
μA(Z)
0
Medium Large
SWRL Rules for Fuzzy Logic Reasoning
MediaTypeSelectionRule-1 ActivityID(?a)∧UsedDevice(?a,?y)∧NetworkBandwidth(?y,"Low") → HasMediaType(?y, Text)
MediaTypeSelectionRule-2 ActivityID(?a)∧UsedDevice(?a, ?y)∧NetworkBandwidth(?y, "Medium")→HasMediaType(?y, Text)∧HasMediaType(?y, Image)
MediaTypeSelectionRule-3 ActivityID(?a)∧UsedDevice(?a, ?y)∧NetworkBandwidth(?y, "High") →HasMediaType(?y,Text)∧ HasMediaType(?y,Image)∧ HasMediaType(?y,Video)
ResourceSizeRule-1ActivityID(?a)∧UsedDevice(?a, ?y)∧ProbLow(?y, ?Tl)∧ProbMedium(?y, ?Tm)∧ProbHigh(?y, ?Th)∧ MaxSize(?y, ?Maxsize) ∧swrlb:multiply(?Lowsize, 0.25, ?Maxsize)∧swrlb:multiply(?Mediumsize, 0.5, ?Maxsize)∧swrlb:multiply(?Largesize, 0.75, ?Maxsize)∧swrlb:multiply(?l, ?Lowsize, ?Tl)∧swrlb:multiply(?m, ?Mediumsize, ?Tm)∧swrlb:multiply(?h, ?Largesize,?Th)∧ swrlb:add(?z1,?l,?m, ?h)∧swrlb:add(?z2, ?Tl, ?Tm, ?Th)∧swrlb:equal(?z2, 1)∧swrlb:divide(?z, ?z1, ?z2) → FileSize(?y, ?z)
AllowedResourceSizeRule-1 ActivityID(?a)∧UsedDevice(?a, ?y)∧FileSize(?y, ?Size)∧ AvailableMemory(?y,?MemorySize)∧ swrlb:lessThan(?MemorySize,?Size) →AllowedSize(?y,?MemorySize)
AllowedResourceSizeRule-2 ActivityID(?a)∧UsedDevice(?a,?y)∧FileSize(?y,?Size)∧AvailableMemory(?y, ?MemorySize)∧ swrlb:greaterThanOrEqual(?MemorySize, ?Size)→AllowedSize(?y,?Size)
81
Other System-Centric Adaptations
• Selecting Resources based on Device Features Language Rule
O/S-Compatibility Rule
82
LanguageRule-1
ActivityID(?a)∧UsedDevice(?a,?y)∧PreferredLanguage(?a, ?z)∧ HasSupportLanguage(?y,?z) → SearchLanguages(?a, ?z)
SystemCentricRule-1
SearchLanguages(?y,?z)∧ExpressedIn(?LR,?z)∧ UsedDevice(?y, ?D)∧ HasMediaType(?D,?b) ∧HapType(?LR,?b)∧HasOS(?D,?c)∧ RunsOn(?LR,?c) → SystemCentric(?y, ?LR)
Learner-Centric Adaptations
Objective: Generate a learning path tailored to the learner’s needs, background, and task at hand.
83
Type of Adaptation Adaptation Process
1. Current task
1 - Get similar knowledge (Isa)
2 – Check for prerequisite knowledge (HasPrerequisite)
2. Learner needs
1- Get core knowledge (HasNecessaryPartOf)
2- If time permits then get related knowledge (HasPartOf)
3. Learning history Infer covered resources 1 - Infer covered concepts (HasCovered)
2 – Infer consumed learning resources (HasConsumed)
Current Task Adaptation• Inference of Similar Learning Knowledge
• Inference of Prerequisite Learning Knowledge
84
SimilarLearningResourceRule-1: ConductedLearningActivity(?L,?a)MakeQuery(?a,?Q)HasKeyword(?Q,?C)IsMappedTo(?C,?LR) → SimilarLR(?a,?LR)
SimilarLearningResourceRule-2: ConductedLearningActivity(?L,?a)MakeQuery(?a,?Q)HasKeyword(?Q,?C)Has(?C,?Ci) ¬Covered(?L,?Ci) IsMappedTo(?Ci,?LRi)¬Consumed(?L,?LRi) → SimilarLR(?a,?LRi)
SimilarLearningResourceRule-3: ConductedLearningActivity(?L,?a)MakeQuery(?a,?Q)HasKeyword(?Q,?C) Isa(?C,?Ci)¬ Covered(?L,?Ci)IsMappedTo(?Ci,?LRi) ¬Consumed(?L,?LRi) → SimilarLR(?a,?LRi)
PrerequisiteLearningResourceRule-1:ConductedLearningActivity(?L,?a)MakeQuery(?a,?Q)HasKeyword(?Q,?C) HasPrerequisite(?Q,?Ci) ¬covered(?L,?Ci) IsMappedTo(?Ci,?LRi) ¬Consumed(?L,?LRi) → PrerequisiteLR(?a,?LRi)
Learner Needs Adaptation
• Inference of Core Learning Knowledge
• Inference of Related (non-core) Learning Knowledge
85
CoreLearningResourceRule-1 : ConductedLearningActivity(?L,?a)MakeQuery(?a,?Q) HasKeyword(?Q,?C) HasNecessaryPart(?Q,?Ci) ¬Covered(?L,?Ci) IsMappedTo(?Ci,?LRi) ¬ Consumed(?L,?LRi) → CoreLR(?a,?LRi)
CoreLearningResourceRule-2 : ConductedLearningActivity(?L, ?a)MakeQuery(?a,?Q)HasKeyword(?Q,?C)IsNecessaryPartOf(?Q,?Ci) ¬Covered(?L,?Ci)IsMappedTo(?Ci,?LRi) ¬Consumed(?L, ?LRi) → CoreLR(?a,?LRi)
NonCoreRelatedLearningResourceRule-1: ConductedLearningActivity(?L,?a)MakeQuery(?a,?Q)HasKeyword(?Q,?C) HasPart(?Q,?Ci) ¬Covered(?L,?Ci) IsMappedTo(?Ci,?LRi) ¬Consumed(?L,?LRi) → NonCoreRelatedLR(?a,?LRi)
NonCoreRelatedLearningResourceRule-2: ConductedLearningActivity(?L,?a) MakeQuery(?a,?Q) HasKeyword(?Q,?C) IsPartOf(?Q,?Ci) ¬Covered(?L,?Ci) IsMappedTo(?Ci,?LRi) ¬Consumed(?L,?LRi) → NonCoreRelatedLR(?a,?LRi)
Example of Learner-Centric Adaptations
Assumptions: - Irene’s query: C++ Loops - Covered Concept by Irene: (Do-While)
Recommended Learning Sequence: - Step 1: Learning Path (Ignoring Learning History) (C++ Loops) → (While) → (Do-While) → (For) → (Looping) - Step 2: Final Learning Path (C++ Loops) → (While) → (For) → (Looping)
86
Similar Concept
Learning Resource
C6
LR6a LR6c
LR37a
LR36a
Non Core Concept
C36 C37
LR45c LR41a
C41 C445
LR1b
C1
LR45a LR1a
LR6b
87
3: Read Ontology
1: Send Query
12: Return
8: Invoke Reasoning
2: Invoke
1: Send Query
4: Infer
Related
Concepts
13: Display Results (WML)
Web Borrower
HTML
WAP Borrower WML
Web Server
Apache-Tomcat-6.0.14
Web Application
Java Servlet
Ecl i pse SDK 3. 3. 1
Jena-2.5.4
OWL Ontology
Global Ontology Space
Protégé 3. 4 Beta
Jess 7
Ontology Reasoning
SWRL-Jess Bridge Java API
Context
Atomic Context XML
9: Aggregate
Context
10: Infer Context
13: Display Results (HTML)
User5: Update Learning
Sequence
7: Save to
Buffer
6: Retrieve
Related LRs
LR-Repository
Learning Resource
XML
Dom4j-1.6.1 Buffer Storage
Retrieved LRs
XML
11: Retrieve
Matched LRs
System Implementation
88
System-Centric Adaptations
89
Learner-Centric Adaptations
90
Learner-Centric Adaptations
Conclusions• Ontologies provide a shared understanding of a domain,
hence allowing semantic interoperability
• Ontologies are useful for improving the accuracy of searches for learning resources that are tailored to the learner’s needs, background, and environment – search engines can look for resources that refer to a precise
learner’s context
• SW provides a learning infrastructure where knowledge, organized in conceptual spaces (based on its meaning) can be semantically queried, discovered, and shared across applications.
Conclusions (2)• SW allows learning applications to share services
without much overhead. Learning services across applications can be integrated by resolving differences in terminology through mappings between ontologies
• SW allows expressing knowledge in well-understood formal semantics
• Automated reasoners can deduce (infer) conclusions from the given knowledge– Logic can be used to uncover ontological knowledge that is
implicitly given – It can also help uncover unexpected relationships and
inconsistencies– Logic can also be used by intelligent agents for making
decisions and selecting courses of learning action
•
Conclusions (3)
• SW provides Web agents with:– Agent communication languages– Formal representation of intentions (negotiation strategies)– Logic to reason based on current facts and negotiation
strategies
• Thus, e-learning systems will use Semantic Web-based Knowledge Systems as key parts of everyday learning cycles
Thank you
Questions?
Acknowledgements & References• RDF Primer , http://www.w3.org/TR/rdf-primer/• OWL, http://www.w3.org/2004/OWL/ • Web Services, http://www.w3.org/2002/ws/ • G. Antoniou & F. Harmelen, A Semantic Web Primer, 2nd Edition
(Cooperative Information Systems) , MIT Press, 2008.• Terence Love et al., The Future of e-Learning: Inclusive learning
objects using RDF.• Andreas Hotho & York Sure, A Short Semantic Web Tutorial,
Knowledge Management Group Institute AIFB, University of Karlsruhe
• Steffen Staab, Querying OWL Ontologies, 2007, http://www.uni-koblenz.de/~staab/Teaching/Tutorials/SSMS-2007/
• Steffen Staab, Ontologies and the Semantic Web, 2006, http://www.uni-koblenz.de/~staab/Teaching/Tutorials/SMBM-2006/103.htm
• Harry Chen, Semantic Web in a Pervasive Context-Aware Architecture, http://ebiquity.umbc.edu/get/a/publication/66.ppt